Changeset 40314 for trunk/ippScripts/scripts/rawcheck_mddb.pl

Timestamp:

Jan 18, 2018, 3:59:12 PM (8 years ago)

Author:

watersc1

Message:

Update to fix problem if MD database contains an object with no linked instances.

File:

• trunk/ippScripts/scripts/rawcheck_mddb.pl (modified) (3 diffs)

trunk/ippScripts/scripts/rawcheck_mddb.pl

@@ -167 +167 @@
         $down_volume{$name} = 1;
     }
+    else {
+        $down_volume{$name} = 0;
+    }
 
     if (($name =~ /ippb05/)||($name =~ /ippb02/)) {
@@ -262 +265 @@
 
         # Get md database results for this exp_id/class_id
-        my $md_query = "select file_id,ins_id,mvol_name,site_name,file,disk_sum,(STRCMP(md5sum,disk_sum) = 0) AS is_valid FROM obj JOIN ins USING(file_id) JOIN vol USING(mvol_id) WHERE exp_id = $exp_id AND class_id = '${class_id}'";
+        my $md_query = "select file_id,ins_id,mvol_name,site_name,file,disk_sum,(STRCMP(md5sum,disk_sum) = 0) AS is_valid FROM obj LEFT JOIN ins USING(file_id) LEFT JOIN vol USING(mvol_id) WHERE exp_id = $exp_id AND class_id = '${class_id}'";
         my $md_instances = $mddb->selectall_arrayref( $md_query );
         
@@ -270 +273 @@
         foreach my $r (@{ $md_instances }) {
             my ($file_id,$ins_id,$vol_name,$site_name,$f,$disk_sum,$is_valid) = @{ $r };
-            $f =~ s%//%/%g;
-            unless (exists($unified_list{$f})) {
-                if (-e $f) {
-                    $unified_list{$f}{exists} = 1;
-                }
-                else {
-                    $unified_list{$f}{exists} = 0;
-                }
-            }
-            # This should only happen if the same scan was ingested twice.
-            # Be paranoid and refuse to cull if this is the case.
-            if (exists($unified_list{$f}{md_ins_id})) {
-                if (($file_id   == $unified_list{$f}{md_file_id})&&
-                    ($vol_name  eq $unified_list{$f}{md_vol})&&
-                    ($site_name eq $unified_list{$f}{md_site})&&
-                    ($disk_sum  eq $unified_list{$f}{disk_sum})&&
-                    ($is_valid  == $unified_list{$f}{is_valid})) {
-                    print "#DUPLICATE INS: $ins_id have $unified_list{$f}{md_ins_id} already\n";
-                    mddb_delete_instance($mddb,$ins_id);
-                    $do_cull = -1;
-                    next;
-                }
-                else {
-                    # The data is inconsistent.  Make a human fix it.
-                    print "INS ERROR: $f -> \n";
-                    print "\t($file_id $ins_id $vol_name $site_name $disk_sum $is_valid)\n";
-                    print "\t($unified_list{$f}{md_file_id} $unified_list{$f}{md_ins_id} $unified_list{$f}{md_vol} $unified_list{$f}{md_site} $unified_list{$f}{disk_sum} $unified_list{$f}{is_valid})\n";
-                    die;
-                }
-            }
-
-            $unified_list{$f}{md_file_id} = $file_id;
-            $unified_list{$f}{md_ins_id}  = $ins_id;
-            $unified_list{$f}{md_vol}     = $vol_name;
-            $unified_list{$f}{md_site}    = $site_name;
-            $unified_list{$f}{disk_sum}   = $disk_sum;
-            $unified_list{$f}{is_valid}   = $is_valid;      
-            
-            if ($md_file_id == -1) {
-                $md_file_id = $file_id;
-            }
-            elsif ($file_id != $md_file_id) {
-                die "Instance $ins_id has bad file_id $file_id expected $md_file_id";
+
+            if ((defined($f))&&(defined($ins_id))) { 
+                $f =~ s%//%/%g;
+                unless (exists($unified_list{$f})) {
+                    if (-e $f) {
+                        $unified_list{$f}{exists} = 1;
+                    }
+                    else {
+                        $unified_list{$f}{exists} = 0;
+                    }
+                }
+                # This should only happen if the same scan was ingested twice.
+                # Be paranoid and refuse to cull if this is the case.
+                if (exists($unified_list{$f}{md_ins_id})) {
+                    if (($file_id   == $unified_list{$f}{md_file_id})&&
+                        ($vol_name  eq $unified_list{$f}{md_vol})&&
+                        ($site_name eq $unified_list{$f}{md_site})&&
+                        ($disk_sum  eq $unified_list{$f}{disk_sum})&&
+                        ($is_valid  == $unified_list{$f}{is_valid})) {
+                        print "#DUPLICATE INS: $ins_id have $unified_list{$f}{md_ins_id} already\n";
+                        mddb_delete_instance($mddb,$ins_id);
+                        $do_cull = -1;
+                        next;
+                    }
+                    else {
+                        # The data is inconsistent.  Make a human fix it.
+                        print "INS ERROR: $f -> \n";
+                        print "\t($file_id $ins_id $vol_name $site_name $disk_sum $is_valid)\n";
+                        print "\t($unified_list{$f}{md_file_id} $unified_list{$f}{md_ins_id} $unified_list{$f}{md_vol} $unified_list{$f}{md_site} $unified_list{$f}{disk_sum} $unified_list{$f}{is_valid})\n";
+                        die;
+                    }
+                }
+                
+                $unified_list{$f}{md_file_id} = $file_id;
+                $unified_list{$f}{md_ins_id}  = $ins_id;
+                $unified_list{$f}{md_vol}     = $vol_name;
+                $unified_list{$f}{md_site}    = $site_name;
+                $unified_list{$f}{disk_sum}   = $disk_sum;
+                $unified_list{$f}{is_valid}   = $is_valid;          
+            }
+
+            if (defined($file_id)) {
+                if ($md_file_id == -1) {
+                    $md_file_id = $file_id;
+                }
+                elsif ($file_id != $md_file_id) {
+                    die "Instance $ins_id has bad file_id $file_id expected $md_file_id";
+                }
             }
         }